US court rules that copyrighted content can be used in AI training, creators face setbacks in protecting their rights

The U.S. District Court for the Northern District of California recently made a key ruling, allowing AI companies to use copyrighted works to train models. This decision has caused a strong shock in the creative industry. Technology media AppleInsider pointed out that the ruling poses a major threat to artists, musicians and writers.

According to reports, Judge William Alsup, while hearing the class action lawsuit of writer Andrea Bartz and others, partially supported the plaintiff's claims, but ultimately ruled that Anthropic's use of pirated materials to train the Claude AI model fell within the scope of "fair use" permitted by law. This means that the common phenomenon that content creators have complained about for many years - AI companies crawling website content and scanning books without permission to train large language models (LLM) - has been recognized at the judicial level.

It is worth noting that these creative achievements that have been obtained in batches are not only used to build the core capabilities of generative AI, but are also transformed into profit tools by commercial companies. However, the original creators of the works have neither received any financial compensation nor lost control of their intellectual property. The three writers' representatives in this case filed a lawsuit in 2024 in an attempt to gain legal protection for the creator community.